ch VDJdb distances between superimposed sequences

Open file with B_CDR3 distances and create plots:

library(reshape2)
library(ggplot2)
library(gplots)

## 
## Attaching package: 'gplots'

## The following object is masked from 'package:stats':
## 
##     lowess

library(plyr)
library(MASS)
library(stringr)

df <- read.csv('seqdata/HomoSapiens/distances_from_pdb_test_Bcdr_3.txt', stringsAsFactors = FALSE, sep='\t')
df$dataset <- paste(df$pdb1, df$pdb2, sep = "_")
plot_imp <- function(imp_dataset) {
  plt <- ggplot(imp_dataset, aes(x=X.1, y=X.2, fill=distance)) +
    geom_tile(data=imp_dataset, 
                   aes(x=X.1, y=X.2, fill=distance), size=2) +
    #geom_tile(data=subset(imp_dataset, distance <= 6), 
    #               aes(x=X.1, y=X.2), size=1, color="black", fill=NA) +
    geom_label(aes(label=paste(AA1, AA2, sep=" ")),size=1.5) +
    scale_fill_gradient2("Distance, A", limits=c(0,18), midpoint = 9,
                       low="#d73027",mid="#e0f3f8",high="#4575b4") +
    labs(title=paste("PDB1: ",imp_dataset$pdb1,"; PDB2: ", imp_dataset$pdb2, sep=""),x=paste(paste("Position in", imp_dataset$pdb1, sep=" "), imp_dataset$pdb1seq, paste(nchar(imp_dataset$pdb1seq[1]), "aminoacids", sep=" "), sep="\n"), y=paste(paste("Position in", imp_dataset$pdb2, sep=" "),imp_dataset$pdb2seq, paste(nchar(imp_dataset$pdb2seq[1]), "aminoacids", sep=" "), sep="\n"))
  print(plt)
}
df.2 <- split(df, f=df$dataset)
for (i in df.2) {
  plot_imp(i)
}

#nothing

Open file with A_CDR3 distances and create plots:

df <- read.csv('seqdata/HomoSapiens/distances_from_pdb_test_Acdr_3.txt', stringsAsFactors = FALSE, sep='\t')
df$dataset <- paste(df$pdb1, df$pdb2, sep = "_")
df.2 <- split(df, f=df$dataset)
for (i in df.2) {
  plot_imp(i)
}

Open file with B_CDR2 distances and create plots:

df <- read.csv('seqdata/HomoSapiens/distances_from_pdb_test_Bcdr_2.txt', stringsAsFactors = FALSE, sep='\t')
df$dataset <- paste(df$pdb1, df$pdb2, sep = "_")
df.2 <- split(df, f=df$dataset)
for (i in df.2) {
  plot_imp(i)
}

Open file with A_CDR2 distances and create plots:

df <- read.csv('seqdata/HomoSapiens/distances_from_pdb_test_Acdr_2.txt', stringsAsFactors = FALSE, sep='\t')
df$dataset <- paste(df$pdb1, df$pdb2, sep = "_")
df.2 <- split(df, f=df$dataset)
for (i in df.2) {
  plot_imp(i)
}

Open file with B_CDR1 distances and create plots:

df <- read.csv('seqdata/HomoSapiens/distances_from_pdb_test_Bcdr_1.txt', stringsAsFactors = FALSE, sep='\t')
df$dataset <- paste(df$pdb1, df$pdb2, sep = "_")
df.2 <- split(df, f=df$dataset)
for (i in df.2) {
  plot_imp(i)
}

Open file with A_CDR1 distances and create plots:

df <- read.csv('seqdata/HomoSapiens/distances_from_pdb_test_Acdr_1.txt', stringsAsFactors = FALSE, sep='\t')
df$dataset <- paste(df$pdb1, df$pdb2, sep = "_")
df.2 <- split(df, f=df$dataset)
for (i in df.2) {
  plot_imp(i)
}